Maybe a manager at each building. They could adjust prices, hire workers. I want to keep it simple though.

State gameplay

Some problems I face now that I've decided to separate states and corporations: The state players need stuff to do, besides paying the military on taxes collected from businesses and citizens on their territory. There's police, healthcare, education, pension. I need to find the simplest solution.

Split corporations

Also, in a real war, could a corporation operate on both sides?

http://en.wikipedia.org/wiki/History_of_General_Motors#World_War_II

By mainstream accounts, General Motors' German subsidiary (Adam Opel AG) was outside the control of the American parent corporation during World War II. Some conspiracy theorists posit that this was a hoax, with the American GM as a secret war profiteer on both sides, but Alfred Sloan's memoir, for example,[16] presents a description of lost control that is much more Occam-compliant than the fringe alternatives. [...] During the war, GM declared it had abandoned its German subsidiary, and took a complete tax write-off worth "approximately $22.7 million", yet after the war, GM collected some $33 million in "war reparations" because the Allies had bombed its German facilities.

Maybe the same player retains control but the finances are managed separately, and are rejoined after the war ends.

Unit production

Factories produce trucks and military units. Trucks can be ordered by any corp. Military units can only be ordered by states from factories on their own land, or bought from other states as arms shipments.

Taxes

Taxes are manufacturer's, wholesale, retail, and income.

http://taxhistory.tax.org/Civilization/Documents/Sales/hst6646/6646-1.htm

Manufacturer's tax would apply to trucks and military products.

Wholesale would apply to uranium, coal, crude oil, "production", and farm products.

Retail would apply to electricity, gasoline and "consumer goods" (food). In real life electricity is probably retail but I should make it wholesale because only buildings use electricity, unless I program some civilian use for it.

Income would apply anywhere labourers are paid.

Autonomous military

I think guns, armour, and vehicles/tanks will be manufactured, but labourers must actually be hired to use them. They would need to be off-duty sometimes to rest and shop/eat.

Murder simulators

I'm tired of the game mechanics of "murder simulators". It's probably why I don't enjoy playing games these days. Most games are in some way about killing enemies, even chess. Games are more and more lacking in motives. But it's still probably a viable solution to game design, with some new mechanics to it. I'm focusing so much on reality though that the focus on war seems hideous sometimes. Maybe a "chibi" style à la Smiley Wars? World in Conflict seems about the proportion I'm aiming for.

There should be some way for corporations to buy out or out-compete/bankrupt other corporations.

Dynamic navmesh

Navmesh navigation would be a lot faster. All you have to do, is split some navmesh polygons whenever a new building is placed. And to use it, just run A* through the polygons and plot the shortest route between connecting shared edges between the path polygons.

How to handle unit avoidance though without making them part of the navmesh? A simple walk-by-on-the-right-side algorithm might do.

Crowds and waves of people would be a marvel to tinker with.

I'm going back to Corporation-States. I want to simulate things like bifurcation, butterfly effect, feedback, chaos theory and cybernetics.

Nothing to do now but wait until this second attempt is operational. In the meanwhile there will be idle thoughts. And noodles.

[edit] A problem I have with the BazaarBot / Ian Parberry model is it appears to require all the bids and asks to occur at the same time in auction.

There's sound cards, video cards, and even a physics processing unit. I wish they made dedicated economics-processing hardware, heh. Something pathfinding-intensive.

I didn't realize there's something called Jump Point Search, an A* pathfinding optimization that might make it 2-30 times faster (the longer the path, the more improvement).